Scene DNA
# From One Image to a Planned Sequence: Your First Scene with Scene DNA
A beautiful image can suggest an entire story. Turning that idea into several shots is where the juggling begins: keeping the location consistent, remembering the lighting, changing the camera, and finding out which video belongs to which shot.
ArtofTomworks Scene DNA brings that planning into one local workspace. It helps you describe what stays the same, decide what changes, and follow supported video generation without wondering what is running in the background.
Imagine a lighthouse at sunset: first the coastline, then the path to the cottage, then its glowing window. Three different views, one shared scene. Here is a practical way to start.
## Choose the package that suits you
Beginners is for your first Scene DNA project, your first storyboard, or your first experience handing a plan to ComfyUI. It also suits experienced creators who prefer a guided introduction. Start with START_HERE_EN.txt, then follow FIRST_PROJECT_EN.md.
Standard suits creators already comfortable with local tools and workflow files. It includes the application, examples and reference guides.
Both editions have exactly the same features. Beginners adds guidance, not restrictions. Neither package includes generation models or automatically sets up your generation environment.
## 1. Give the scene its shared identity
Describe the location, subject, appearance, time of day, lighting, palette and visual style in the shared scene recipe. Add a reference image if you have one.
Keep the details that should remain consistent locked. Every shot inherits them; unlock a field when you deliberately want a difference. This keeps the instructions organized, although visual consistency still depends on your generator.
## 2. Plan a small sequence
Start with three editable shot templates. Give each a clear purpose: establish the place, follow an action, reveal a detail. Adjust framing, camera movement, action and intended duration. The larger composition view helps you explore framing before committing to a prompt.
There are separate copy buttons for image prompts and video prompts. Optional local AI can suggest three editable video-scene ideas from a reference image; review and apply the one you want. Image analysis needs running Ollama and an installed vision-capable model. Basic planning works without AI.
## 3. Copy prompts—or prepare supported shots
For planning alone, extract the ZIP and use START_OFFLINE.cmd. For optional local AI and the shot runner, install Node.js 22+ separately and use START.cmd, keeping its console open.
Copied prompts can be used with your preferred generator. The integrated runner currently supports recognized H3 Simple/Pro ComfyUI workflows. Load a compatible workflow source and choose its saved-video output. If using SwarmUI, connect to its actual ComfyUI backend.
Check each shot’s start image, orientation, dimensions and duration. Begin with a modest-resolution short test, then choose the full planned length when satisfied. The displayed frame-derived duration may differ slightly from the requested seconds.
Prepare first, then explicitly start generation. You can run selected shots sequentially or save individual workflow JSON files for manual use. Imported model, LoRA, sampler and seed settings are retained in execution copies; reimport the source if you change those settings elsewhere.
## 4. Follow the work and find the results
Stay in Scene DNA’s shot monitor to see the active shot, actual settings, progress and available backend previews. Enlarge the view when needed. Some loading or decoding stages provide no preview image; the status still matters. The separate ComfyUI canvas is not automatically replaced by the submitted workflow.
Completed videos are saved under the backend’s configured output folder, normally within ArtofTomworks/SceneDNA/. Each result shows its saved path and a download option. Downloading creates an additional copy.
A useful detail: completed, unchanged shots can be reused from earlier run folders. Five completed shots may therefore mean four new files plus one existing result elsewhere. They are not copied into the newest folder. Check the result cards before assuming something is missing or generating it again.
## Keep the next session easy
Save a Scene DNA project file to preserve your plan and image thumbnails. This opens in Scene DNA; an exported workflow JSON opens in ComfyUI. They serve different purposes. You can also export a prompt sheet or print a storyboard/PDF.
Scene DNA 0.7.0 is a public beta. It plans and coordinates; it does not automatically join the shots into a finished film. Sound directions guide prompts rather than generating audio. Avoid running heavy local image analysis alongside GPU-intensive generation on a memory-limited machine.
The software is free for personal and commercial creative use under the included license; resale or paid bundling of the software is not permitted. Generation models and third-party tools have their own terms.
[Get Scene DNA, both editions and the guides on GitHub](https://github.com/artoftomworks-debug/artoftomworks-scene-dna).
Try one scene and three shots. Then tell me which part made your work easier—and which button still made you stop and think. That feedback helps shape what comes next.
Cover: AI-generated promotional illustration, not a software screenshot.